Valvano, Nho and Anderson May 7, 1999 Page 1

Transcription

1 Valvano, Nho and Anderson May 7, 1999 Page 1 Analysis of the Weinbaum-Jiji Model of Blood Flow in the Canine Kidney Cortex for Self-Heated Thermistors Jonathan W. Valvano*, Sungwoo Nho* and Gary T. Anderson *Biomedical Engineering Program The University of Texas at Austin Department of Electronics and Instrumentation University of Arkansas at Little Rock ABSTRACT The Weinbaum-Jiji equation can be applied to situations where: 1) the vascular anatomy is known; 2) the blood velocities are known; 3) the effective modeling volume includes many vessels; and 4) the vessel equilibration length is small compared to the actual length of the vessel. These criteria are satisfied in the situation where steady-state heated thermistors are placed in the kidney cortex. In this paper, the Weinbaum-Jiji bioheat equation is used to analyze the steady state response of four different sized self-heated thermistors in the canine kidney. This heat transfer model is developed based on actual physical measurements of the vasculature of the canine kidney cortex. In this model, parallel-structured interlobular arterioles and venules with a 60 µm diameter play the dominant role in the heat transfer due to blood flow. Continuous power is applied to the thermistor, and the instrument measures the resulting steady state temperature rise. If an accurate thermal model is available, perfusion can be calculated from these steady-state measurements. The finite element simulations correlate well in shape and amplitude with experimental results in the canine kidney. In addition, this paper shows that the Weinbaum- Jiji equation can not be used to model the transient response of the thermistor because the modeling volume does not include enough vessels and the vessel equilibration length is not small compared to the actual length of the vessel.

2 Valvano, Nho and Anderson May 7, 1999 Page 2 NOMENCLATURE A a B C c D f g k l l i, l j L e L steady state applied thermistor power (12 mw) interlobular vessel radius (cm) sinusoidal applied thermistor power (8 mw) steady state temperature rise in thermistor ( C) specific heat (mw-sec/g- C) sinusoidal temperature rise in thermistor ( C) frequency (0.05 Hz) empirically determined calibration coefficients thermal conductivity (mw/cm- C) distance from arteriole to venule (0.08 cm) direction cosines vessel equilibration length (cm) vessel length (1 cm) M kidney cortex mass (50 g) n density of vessels in kidney cortex (1/cm 2 ) P(t) Pe r,θ,z total power applied to thermistor (mw) Peclet number, 2 ρb cb a v / kb cylindrical coordinates (cm,radians,cm) S kidney cortex cross sectional area (50 cm 2 ) T t v temperature rise T - T0 ( C) time (sec) velocity of blood (cm/s) V thermistor volume (cm 3 ) w x i tissue perfusion (ml/100g-min) coordinate axis

3 Valvano, Nho and Anderson May 7, 1999 Page 3 α T ε φ thermal diffusivity (cm 2 /sec) volumetric average thermistor temperature rise ( C) L e /L normalized vessel equilibration length phase between power and temperature (radians) ρ density (g/cm 3 ) σ Σ coupling shape factor between tissue/vessels (2.5) σ coupling shape factor between two vessels (1.8) Subscripts a b eff g h m interlobular arterioles blood (saline perfusate) effective or enhanced (Weinbaum-Jiji) glass shell around the thermistor measured in the heated thermistor tissue medium (intrinsic) 0 initial, measured in the sensing thermistor p sh sin ss v w thermistor probe glass shell from sinusoidally heated thermistor from steady-state heated thermistor interlobular venules wire leads of the thermistor

4 Valvano, Nho and Anderson May 7, 1999 Page 4 INTRODUCTION There has been considerable interest in developing more accurate thermal models that describe blood flow in biologic tissue. This paper demonstrates how to apply the Weinbaum-Jiji (W-J) model to a specific bioheat transfer problem. One starts with actual physical parameters of the tissue, and applies them to test the necessary assumptions which allow the use of the W-J model. Then, one calculates the specific W-J model parameters using actual physical measurements. Although the particular problem analyzed in this paper is self-heated thermistors in the canine kidney cortex, the far-reaching impact of this paper lies with the general process concerning how to develop and apply the W-J model. One of the requirements necessary for self-heated thermistors to be able to quantify perfusion is the availability of an accurate model describing the heat transfer due to blood flow. In this paper, the response of self-heated thermistors has been numerically evaluated using the enhanced thermal conductivity model of W-J in the canine kidney cortex. Physical parameters for the W-J equation were measured directly from vascular casts of the canine kidney cortex. The majority of heat transfer in the kidney cortex occurs in the interlobular arteriole-venule pairs. The finite element simulations correlate well in shape and amplitude with experimental results in the canine kidney. In addition, this paper shows that the W-J equation can not be used to model the transient response of the thermistor because the modeling volume does not include enough vessels and the vessel equilibration length is not small compared to the actual length of the vessel. There is considerable interest in thermal models which describe the heat transfer due to perfusion (Chato 1990). The W-J equation has been applied to peripheral tissue and arm (Song 1987, Charny 1990). Xu (1991) used the W-J model to theoretically examined the overall heat transfer due to blood flow in the pig renal cortex. Experimental results for selfheated thermistors have been inconclusive as to which thermal model best describes perfusion (Valvano 1984, Patel 1987, Anderson 1992, Anderson 1993). This paper

5 Valvano, Nho and Anderson May 7, 1999 Page 5 examines the specific case of self-heated thermistor measurements of perfusion in the canine kidney cortex, which has vascular dimensions considerably different from the pig kidney. A numerical model is developed that is based on the vasculature of the canine kidney cortex. This numerical model is then used to explain the effects of perfusion on a thermistor that is heated with steady state power. Four issues must be considered when modeling heat transfer in tissue. First, one must have a clear concept of the goal of the model. In this paper, the model is used to develop a relationship between self-heated thermistor measurements and perfusion in the kidney cortex. Second, the model must carefully adhere to realistic anatomy and physiology. One should make simplifying assumptions because they are realistic, and not just because they allow a solution to be found. Third, before formulating a model, it is important to define the volume of interest. A model which accurately describes whole body temperature will not be appropriate to study heat transfer across cell membranes. The length scale of the model is determined by the steady state temperature field around a thermistor. The temperature field around a 0.15 cm diameter thermistor extends to about 0.5 cm 3. Fourth, it is critical to verify the model with carefully acquired experimental data. SELF-HEATED THERMISTOR TECHNIQUE Although tissue perfusion is believed to be an important factor in many medical conditions such as heart disease, vascular surgery, plastic surgery, transplants, and cancer therapy, there is currently no widely accepted method to quantify perfusion for a majority of applications. The basic concept for the thermal diffusion probe was introduced by Chato (1968), and developed by Balasubramaniam (1977), Jain (1979), Patera (1979), Holmes (1980), Valvano (1984,1987), Patel (1987), and Anderson (1992.) A thermistor is placed into a perfused tissue and a microcomputer based instrument is used to heat the thermistor with a predetermined power. The resultant temperature rise in the thermistor is then

6 Valvano, Nho and Anderson May 7, 1999 Page 6 measured by the instrument. Both tissue thermal conduction and perfusion act to carry heat away from the thermistor. From the steady-state temperature rise in the thermistor, an effective thermal conductivity of tissue, k ss, is calculated. From the k ss calculation, perfusion can be extracted based on the knowledge of the intrinsic tissue conductivity and of the particular perfusion model that is used. The self-heated thermistor technique currently has the following limitations: Large probes cause trauma when inserted; Small probes have a small effective measurement volume where only a few vessels cross the temperature field; Knowledge of k m is required to detect the situation where there is no flow in the tissue; An accurate model is required to quantify perfusion. Although noninvasive surface probes produce no trauma, they are unreliable due to poor contact with tissue and uncertain boundary conditions at the contact surface (Patel 1987.) The use of ultrasonic heating (Newman 1985 and Anderson 1990) can reduce the effect of trauma, but still requires accurate knowledge of tissue properties, heating pattern, and perfusion model. Since the intrinsic thermal conductivity affects heat transfer, its value should be known prior to the perfusion measurement or k m must be measured as part of the experimental procedure. Holmes and Chen (1980) solved this problem using pulse decay method. By using a repetitive string of heating pulses, they are able to determine perfusion without a priori knowledge of k m. Bowman (Balasubramaniam 1977, Valvano 1984, Bowman 1989) heats the thermistor with a feedback circuit which maintains the probe at a constant temperature, and uses the initial power response to estimate k m. Another approach which measures intrinsic conductivity in the presence of perfusion uses sinusoidal heating (Valvano 1987, Anderson 1992.) This technique uses simultaneously applied steady state and sinusoidal power. The power

7 Valvano, Nho and Anderson May 7, 1999 Page 7 P(t) = A + B sin(2πft) = sin(2π 0.05 t) mw (1) is applied to the heat thermistor. The resulting temperature, T h (t), in the heat thermistor is measured. A sense thermistor about 0.5 cm away measures the tissue baseline temperature, T 0. The microcomputer-based instrument calculates, T(t)=T h (t)-t 0. Nonlinear regression is used to determine C,D, and φ from the resulting thermistor temperature: T(t) = C + D sin(2πft+φ) (2) Experimentally it has been shown that the steady state response is affected by both conduction and perfusion whereas if the frequency is fast enough the perfusion has a negligible effect on the sinusoidal response. Thus, effective and intrinsic conductivity s are calculated from the following equations: Effective: 1 k ss = g 1 + g 2 C/A (3) Intrinsic: 1 k sin = g 3 + g 4 D/B (4) This approach was numerically verified for the spherical thermistor (Anderson 1992.) Experimental data show that the phase shift is also sensitive to perfusion, but not as reliable as the steady state response k ss. Effective: 1 k φ = g 5 + g 6 tan φ (5) Coefficients g 1 -g 6 are determined by operating the probe in two unperfused media of known thermal properties. In the absence of perfusion, k ss k sin and k φ will be equal. The

8 Valvano, Nho and Anderson May 7, 1999 Page 8 measurements in agar-gelled water and glycerol give two sets of C/A,D/B,φ data. Since the thermal conductivities of agar-gelled water and glycerol are known, this yields six equations (two sets of Eqs. 3,4,5) and six unknowns (g 1 -g 6.) VASCULAR ANATOMY OF THE KIDNEY CORTEX Vascular casts of the canine kidney cortex were obtained to get the numerical data required to construct the perfusion model. Perfusion rates in the kidney are among the highest found in mammalian organs due to its filtration and reabsorption functions. Blood proceeds from relatively large arcuate arteries located in the cortical-medullary junction to small interlobular arteries located in the cortex to the glomeruli. The glomeruli function to retain cells in the circulation and also allow filtration of fluid into the tubular space. Three canine kidneys were excised and the renal arteries and veins were cannulated. The kidneys were cleared with a mannitol-saline perfusate for 30 minutes. Degassed microfil silicone rubber (Canton Bio-medical Products, Boulder, CO) was slowly injected over a period of about 30 seconds. The first kidney cast was made with both the arteries and veins by injecting 15 ml into the renal artery, and stopping when the silicone appeared in the renal vein. The second cast was made of just the arteries by injecting 8 ml into the renal artery. The last cast was made of just the veins by injecting 8 ml into the renal vein. After injection, the kidneys were refrigerated for 24 hours. 1 mm slices were made at various angles. The tissue slices were soaked in increasing concentrations (25%, 50%, 75%, 95%, 100%) of ethyl alcohol each for 24 hours. A methyl salicylate soak was then used to clear the tissue from the casts. Experimentally the thermistors are located in the center of the cortex which is about 0.5 cm from the surface. Therefore, it is at this kidney location that the vascular statistics were collected. Figure 1 is a typical r-z section showing the interlobular venules. The major heat carrying vessels in the cortex are the interlobular arterioles and venules. These vessels run predominantly perpendicular to the kidney surface with only occasional branching. Figure 2

9 Valvano, Nho and Anderson May 7, 1999 Page 9 is a typical r-θ section which also shows the interlobular venules. Figures 3 and 4 show the arterial system for the canine kidney cortex. The visual acuity in the venous casts is superior to the arterial casts because the microfil silicone rubber injected in the vein does not fill the glomeruli. Photographs like Figs. 2 and 4 were used to count the vessel density. There were 34 interlobular arterioles counted in an area of 0.58 cm 2 (n a 60 interlobular arterioles/cm 2.) There were 313 interlobular venules counted in an area of 4.12 cm 2 (n v 82 interlobular venules/cm 2.) Interlobular arterioles have a radius, a, of about cm in the center of the cortex and a length of about 1 cm. The majority of artery-vein countercurrent pairs exist in the cortex near the medulla. In the center of the cortex, there are few artery-vein countercurrent pairs. The interlobular arterioles and veins are typically at least several vessel diameters apart. The fluid velocity can be estimated from the total kidney perfusion, w. The vessel density, n, is taken as the average of n a and n v. The 6000 is necessary to convert the perfusion units ml/100g-min into the velocity units of cm/sec. v = w M 6000 π a 2 S n a+n v w (6) The Peclet number can be calculated from the velocity. Pe = 2 a v α b 4 v w (7) c b Specific heat 4000 mw-sec/g- C k b Thermal conductivity 6 mw/cm- C α b Thermal diffusivity cm 2 /sec ρ b Density 1 g/cm 3 Table 1. Parameters of the saline perfusate.

10 Valvano, Nho and Anderson May 7, 1999 Page 10 a Interlobular vessel radius cm k m Tissue thermal conductivity 5 mw/cm- C l Arteriole/venule distance 0.08 cm L Length of the vessel 1 cm M Kidney cortex mass 50 g n Vessel density 71 vessels/cm 2 S Surface area at center of the cortex 50 cm 2 α m Tissue thermal diffusivity cm 2 /sec Table 2. Typical parameters of the alcohol-fixed kidney cortex. APPLICATION OF THE WEINBAUM-JIJI EQUATION The W-J equation can be used only if equilibration length is smaller than the actual vessel length (Baish 1986, Qi 1990.) From Eq. (6), a perfusion of 100 ml/100g-min results in a velocity of about 8.3 cm/sec. ε = L e L = π a 2 ρ b c b v k m σ σ Σ L 0.09 (8) Since ε is much less than 1, the W-J model can be applied to the canine kidney cortex. A two dimensional axisymmetric finite element code was used to model the effects of the interlobular arteries and veins on a self-heated thermistor probe. The complete element grid is shown in Fig. 5. Four different thermistors were evaluated, and the close-up showing one thermistor can be found in Fig. 6. Table 3 contains the dimensions of the four thermistors.

11 Valvano, Nho and Anderson May 7, 1999 Page 11 Wire Diameter mm Shaft Diameter mm Probe Radii mm Bead Size mm mm k p mw/cm- C α p cm 2 /sec Table 3. Specifications for the four thermistor probes. Neglecting the small amount of urine flow, the arterial flow must equal the venous flow. Even though there were slightly more veins than arteries, we assigned the size and density of both the arteries and veins to be the average of the measured values. Seventy one (the average of 60 and 82) artery-vein pairs per cm 2 were assumed to be equally spaced. The W-J equation can be found in their 1985 paper: T x i (k ij ) eff x j - π2 na 2 k b 2 4σk m Pe 2 l j l i x i T x j = ρ m c m T t (9) where (k ij ) eff = k m (δ ij + π 2 4σk m 2 n a2 k b 2 Pe 2 l i l j ) (10) l i and l j are direction cosines and δ ij is the Kronecker delta function. The affect of the local capillary bleed-off is automatically included in the effective thermal conductivity (Weinbaum 1985.) In the canine kidney cortex, the interlobular arteries and veins run parallel to one another, and in one direction only, making l- i / x i = 0. Equation (9) can therefore be rewritten in cylindrical coordinates as: 1 r r T r k m r + z T k eff z = ρ mc m T t (11) The enhanced conductivity, which occurs only in the z direction, becomes

12 Valvano, Nho and Anderson May 7, 1999 Page 12 k eff = k m + π2 4 σ k m n a 2 k b 2 Pe w 2 (12) where σ is given by π σ = cosh -1 l 2a (13) The directional cosine terms have been dropped because of the parallel vessel structure in the kidney cortex. Because the major function of the kidney is filtration, the metabolic heat term is small compared to the flow term. In the alcohol-fixed kidney there is no metabolism. Thus, the metabolic heat term is also neglected. Power is applied uniformly within the thermistor material: 1 r r T r k p r + z T k p z + P(t) V = ρ pc p T t (14) Passive conduction occurs in the glass shell and in the wire leads 1 r r T r k sh r + z T k sh z = ρ shc sh T t (15) 1 r r T r k w r + z T k w z = ρ wc w T t (16) Continuity of flux and temperature are maintained by the finite element code at the material interfaces. The analytic steady state solution of a finite-size spherical thermistor placed in homogeneous infinite-size unperfused tissue has a 1/r tissue temperature dependence (Balasubramaniam 1977)

13 Valvano, Nho and Anderson May 7, 1999 Page 13 T(r) = A 4πk m r (17) where A is the applied steady state thermistor power and r is in spherical coordinates. A remote boundary condition simulates this 1/r temperature dependence: r T n + T = 0 (18) where r in this equation is in spherical coordinates, and n is the normal vector to the surface boundary. The mixed boundary condition of Eq. (18) was evaluated for the situation in which an analytic solution exists (Valvano 1985.) Comparisons between numerical and analytical results show that Eq. (18) has significantly less error than either the constant temperature boundary condition (which underestimates temperature) or the constant flux boundary condition (which overestimates temperature.) The arteriole and venule inlet temperatures are assumed to be T 0. Experimentally this is obtained by placing the alcohol-fixed kidney in a temperature controlled water bath. Thus, the only temperature perturbation is caused by the self-heated thermistor. The temperature in the thermistor is volume averaged to calculate T(t). The T(t) numerical data from 300 to 318 seconds was fit to Eq. (2) to determine C, D, and φ. Coefficients g 1 through g 4 were determined for the numerical simulation using a procedure similar to the experimental calibration. The model was run using two thermal conductivities, k m equal to 4 and 8, and g 1 through g 4 were calculated from Eqs. (3) and (4). RESULTS The steady state and sinusoidal temperatures of the thermistor with a 0.08 radius (Fig. 6) are plotted versus distance from the thermistor in Fig. 7. The size of this thermistor

14 Valvano, Nho and Anderson May 7, 1999 Page 14 approximated the one used in the actual kidney experiment. The effective measurement volume includes tissue with a temperature rise above 0.1 C. Table 4 shows that the steady state field includes many vessels, while the sinusoidal field contains only a few. The sinusoidal temperature field increases as the excitation frequency decreases (Valvano 1984, Valvano 1987.) The fact the sinusoidal temperature field only crosses about 4 arterioles and 6 venules and that ε is 0.3 dictates that the W-J equation can not be used to study the sinusoidal response. Indeed, it is this small measurement volume which allows the sinusoidal response to predict k m in the presence of perfusion. A model which incorporates explicit vessels (e.g., the small artery model of Anderson 1993) must be used to simulate the transient response of the thermistor. Symbol Steady state Sinusoidal Distance to 0.1 C r 0.5 cm 0.15 cm Length scale L t = 2r 1cm 0.3 cm Equilibration length L e /L Equilibration length L e /L t Effective volume 4/3 π r cm cm 3 Surface area crossed π r cm cm 2 Vessels crossed (n a +n v ) π r Table 4. Equilibration length and measurement volume. Figure 8 plots effective conductivity, k ss, as calculated by Eq. (3) for both the FEM simulation and an actual canine kidney experiment. The FEM simulation is similar in shape and magnitude with actual experimental data. A number of potential factors might explain the differences: The thermistor model has one wire and uniform applied power; The thickness of the glass layer was approximated;

15 Valvano, Nho and Anderson May 7, 1999 Page 15 The thermal diffusivity of the bead was derived from simple lumped thermal experiments; The vascular statistics {n,l,a} were measured in just 3 kidneys; σ was calculated using the equation for a single vessel pair. SENSITIVITY ANALYSIS Due to the accuracy of the computer control system the power measurement errors are small compared to the temperature measurement errors. The sensitivity of a transducer is defined as the change in heated thermistor temperature that occurs with a change in perfusion: C C w=w1 - C w=w2 w = w 1 - w 2 = C ml/100g-min (19) The sensitivity increases with applied power. In order to avoid tissue damage, one must limit the probe average temperature rise (C in Eq. 2) to 4 C. The maximum power (A in Eq. 1) that can be applied to a thermistor increases linearly with probe size (Balasubramaniam 1977.) Given this restriction on the maximum probe temperature, the steady state conductivity sensitivity ( C/ k) is independent of probe size. The four thermistors described in Table 3 were used to numerically evaluate the sensitivity of the thermistor to perfusion. Applied thermistor power A was varied to make C at no flow exactly 4 C. Figure 9 plots the perfusion sensitivity versus thermistor size. The smaller thermistors are slightly more sensitive to perfusion than the larger probes. If the Pennes Equation were used the smaller thermistors would have been much less sensitive to perfusion than the larger probes (Valvano 1984.) The sensitivity increases with perfusion for the W-J Equation, and decreases with perfusion for the Pennes Equation. Thus, it is very important to choose a realistic bioheat transfer model.

16 Valvano, Nho and Anderson May 7, 1999 Page 16 Due to the 12 bit A/D, there is a fundamental limitation in the system s ability to measure temperature and power. The perfusion resolution, δw, is defined as the smallest change in perfusion that can be detected by the instrument. Because the power is known accurately, the perfusion resolution is limited by the temperature resolution, δt: δw = δt C/ w (20) The temperature resolution, δt, in this system is about C. For all four probes the resolution was below 4 ml/100g-min. DISCUSSION The W-J bio-heat equation was used to analyze the steady state response of selfheated thermistors in the canine kidney. This heat transfer model was developed based on actual physical measurements of the vasculature of the canine kidney cortex. In this model interlobular arteries and veins with a 60 µm diameter play a dominant role in the heat transfer due to blood flow. The instrument measures effective conductivity, k ss, using Eq. (3). In order to measure perfusion using the self-heated thermistor, a relationship between perfusion and k ss must exist. The finite element model of the W-J equation is useful for suggesting the proper shape of the k ss versus perfusion response. Because of the uncertainties in the thermistor and biologic tissue, an empirical perfusion calibration will be required. Experimental results suggest a simple linear response: w = g 7 (k ss - k m ) (21) The finite element model of the W-J equation suggests a more complex response: w = g 7 (k ss - k m ) g 8 (22)

17 Valvano, Nho and Anderson May 7, 1999 Page 17 where g 7 and g 8 are determined empirically by operating the probe in a tissue with known perfusion. The finite element results were fit to Eq. (22), and the resulting errors were less than 3 ml/100g-min. With this empirical calibration procedure, the instrument is calibrated in one tissue and used in a different tissue. Errors due to tissue variability can be minimized by calibrating in tissue which closely resembles the ultimate in vivo tissue. The fact that only the steady state response varies with blood perfusion is convincing evidence that there is little blood-tissue heat transfer in the smaller vessels. Conversely, if the small vessels which exist in the sinusoidal measuring volume did contribute to bloodtissue heat transfer, then k sin would be sensitive to perfusion. These observations also confirm predictions by Charny (1990) that the blood-perfusion term in Pennes equation does not describe heat exchange in the small vessels as nearly everyone has assumed. CONCLUSIONS This paper describes a procedure which begins with physical measurements of the tissue vasculature, and develops the Weinbaum-Jiji model parameters for the canine kidney cortex. The Weinbaum-Jiji model can be used to model steady state self-heated thermistors because the equilibration length is smaller than the actual vessel length. Because the temperature field generated by the sinusoidal heating was too small, the Weinbaum-Jiji model could not be used to study the transient behavior of the thermistors. Nevertheless, steady state finite element simulations were used to solve the Weinbaum-Jiji equations. These numerical results correlated well in shape and amplitude with actual experimental measurements made in the canine kidney. ACKNOWLEDGMENTS This research was funded in part by grants from National Institutes of Health GM A3, The Whitaker Foundation, and The Texas Advanced Technology Program #

18 Valvano, Nho and Anderson May 7, 1999 Page 18 REFERENCES Anderson, G.T., Burnside, G., 1990, A Noninvasive Technique to Measure Perfusion using a Focused Ultrasound Heating Source and a Tissue Surface Temperature Measurement, ASME Winter Annual Mtg., Dallas, HTD. Vol. 147, pp Anderson, G.T., Valvano, J.W. and Santos, R.R., 1992, Self-Heated Thermistor Measurements of Perfusion, IEEE Trans. on Biomed. Eng., Vol. 39, No. 9, pp Anderson, G.T., Valvano, J.W., 1993 A Small Artery Heat Transfer Model for Self-Heated Thermistor Measurements of Perfusion in the Kidney Cortex, J. of Biomech. Eng., in press. Baish, J.W., Ayyaswamy, P.S., and Foster, K.R., 1986, Small-Scale Temperature Fluctuations in Perfused Tissue During Local Hyperthermia, J. of Biomech. Eng., Vol. 108, pp Baish, J.W., Ayyaswamy, P.S., and Foster, K.R., 1986, Heat Transport Mechanisms in Vascular Tissues: A Model Comparison, J. of Biomech. Eng., Vol. 108, pp Balasubramaniam, T.A., Bowman, H.F., 1977, Thermal Conductivity and Thermal Diffusivity of Biomaterials: A Simultaneous Measurement Technique, J. of Biomech. Eng., Vol. 99, pp Bowman, H.F., et al., 1989, Use of Effective Conductivity for Hyperthermia Treatment Planning, ASME Winter Annual Mtg., San Francisco, HTD Vol. 126, BED Vol. 12, pp Charny, C.K., Weinbaum, S., and Levin, R.L., 1990, An Evaluation of the Weinbaum-Jiji Bioheat Equation for Normal and Hyperthermic Conditions, J. of Biomech. Eng., Vol. 112, pp

19 Valvano, Nho and Anderson May 7, 1999 Page 19 Chato, J.C., 1968, A Method for the Measurement of Thermal Properties of Biologic Materials, Symposium on Thermal Problems in Biotechnology, ASME, N.Y, LCN , pp Chato, J.C., 1980, Heat Transfer to Blood Vessels", J. of Biomech. Eng., Vol. 102, pp Chen, M.M., et al., 1980, Microvascular Contribution to Tissue Heat Transfer, Annals of the NY Acad. of Sci., Vol. 335, pp Chato, J.C., Oleson, J.R., 1990, Biothermal Engineering, Report on a Workshop Held at Allerton House of the University of Illinois at Urbana-Champaign, April Holmes, K.R., Chen, M.M., 1980, In vivo Tissue Thermal Conductivity and Local Blood Perfusion Measured with Heat Pulse-decay Method, Advances in Bioeng., pp Jain, R.K., 1979, Transient Temperature Distributions in an Infinite, Perfused Medium due to a Time-Dependent, Spherical Heat Source, J. of Biomech. Eng., Vol. 101, pp Newman, W.H., Lele, P.P., 1985, A Transient Heating Technique for the Measurement of Thermal Properties of Perfused Biological Tissue, J. of Biomech. Eng., Vol. 107, pp Patel, P.A., Valvano, J.W., Pearce, J.A., Prahl, S.A., and Denham, C.R., 1987, A Self-Heated Thermistor Technique to Measure Effective Thermal Properties from the Tissue Surface, J. of Biomech. Eng., Vol. 109, pp Patera, et al., 1979, Prediction of Tissue Perfusion from Measurement of the Phase Shift between Heat Flux and Temperature, ASME Winter Annual Mtg., Paper #79- WA/HT-71. Pennes, H.H., 1948, Analysis of Tissue and Arterial Blood Temperature in the Resting Human Forearm, J. of Applied Physiology, Vol. 1, pp

20 Valvano, Nho and Anderson May 7, 1999 Page 20 Qi, Y., Wissler, E.H., 1990, A Three-dimensional, Finite Element Analysis of Countercurrent Heat Transfer Between an Artery-Vein Pair and the Surrounding Tissue Cylinder, ASME Winter Annual Mtg., Dallas, HTD Vol. 147, pp Song, W.J., Weinbaum, S., and Jiji, L.M., 1987, A Theoretical Model for Peripheral Tissue Heat Transfer Using the Bioheat Equation of Weinbaum and Jiji, J. of Biomech. Eng., Vol. 109, pp Valvano, J.W., et al., 1984, The Simultaneous Measurement of Thermal Conductivity, Thermal Diffusivity and Perfusion in Small Volume of Tissue, J. of Biomech. Eng., Vol. 106, pp Valvano, J.W., Hayes, L.J., 1985, Steady State Analysis of Self-Heated Thermistors Using Finite Elements, J. of Biomech. Eng., Vol. 107, pp Valvano, J.W., Badeau, A.F., and Pearce, J.A., 1987, Simultaneous Measurement of Intrinsic and Effective Thermal Conductivity, Heat Transfer in Bioengineering and Medicine, ASME Winter Annual Mtg., Boston, HTD Vol. 95, BED Vol. 7, ed. Chato, Diller, Diller, Roemer, pp Weinbaum, S., and Jiji, L., 1985, A New Simplified Bioheat Equation for the Effect of Blood Flow on Average Tissue Temperature, J. of Biomech. Eng., Vol. 107, pp Wissler, E.H., 1988, An Analytical Solution Countercurrent Heat Transfer Between Parallel Vessels with a Linear Temperature Gradient", J. of Biomech. Eng., Vol. 110, pp Xu, L.X., Chen, M.M., Holmes, K.R., and Arkin, H., 1991, The Theoretical Evaluation of the Pennes, the Chen-Holmes and the Weinbaum-Jiji Bioheat Transfer Models in the Pig Renal Cortex, ASME Winter Annual Meeting, Atlanta, HTD, Vol. 189, pp

21 Valvano, Nho and Anderson May 7, 1999 Page 21 List of Figures Figure 1. Vascular cast of the veins in a canine kidney cortex (r,z plane). Figure 2. Vascular cast of the veins in a canine kidney cortex (r, θ plane). Figure 3. Vascular cast of the arteries in a canine kidney cortex (r,z plane). Figure 4. Vascular cast of the arteries in a canine kidney cortex (r, θ plane). Figure 5. Complete finite element grid. Figure 6. Close-up of the finite element grid modeling a thermistor with a bead radius of cm. Figure 7. Steady state and sinusoidal temperature response versus radius distance. Figure 8. Steady state thermal conductivity versus perfusion. Figure 9. Perfusion sensitivity versus thermistor size.